We consider a special class of non-convex programs that are encountered in dispatching the generating units of an electric company. The general form of these problems is $\{\min c^Tx: l^j \le A^jx^j\le u^j, x^j\ge 0, \sum_{j=1}^mx_{i,t}^j=f_i(y_{i,t}), By=d, y\ge 0\}$, where $c\ge 0$, all entries in $A$ are non-negative, and $f_i(y_{i,t})\ge 0$. We show that one can achieve global optimality in the case of $l^j=0$. For the special case in which $A^j$ is a vector of ones, we suggest an alternative formulation in which the nonlinearity is moved to the objective function. Numerical results indicate significant improvement in the number of iterations and computer time needed to solve the problem.

By:* S. Takriti and B. Krasenbrink *

Published in: European Journal of Operational Research, volume 112, (no 2), pages 460-6 in 1999

Please obtain a copy of this paper from your local library. IBM cannot distribute this paper externally.

Questions about this service can be mailed to reports@us.ibm.com .